( Academy of Management Journal
2001, Vol. 44, No. 5, 956-974.
THE INFLUENCE OF DEMOGRAPHIC HETEROGENEITY ON
THE EMERGENCE AND CONSEQUENCES OF COOPERATIVE
NORMS IN WORK TEAMS
JENNIFER A. CHATMAN
University of California, Berkeley
FRANCISJ. FLYNN
Columbia University
Drawing from social categorization theory, we found that greater demographic heterogeneity led to group norms emphasizing lower cooperation among student teams and
officers from ten business units of a financial services firm. This effect faded over time.
Perceptions of team norms among those more demographically different from their
work group changed more, becoming more cooperative, as a function of contact with
other members. Finally, cooperative norms mediated the relationship between group
composition and work outcomes.
increased contact on cooperative norms for demographically similar and different people. Finally,
we examine how inconsistencies in the relationship between group heterogeneity and work outcomes might be explained by considering the mediating role of norms. Thus, this study may explain
the contradictory findings described above; the
negative effects of demographic heterogeneity may
diminish when norms that encourage a focus on
interdependent objectives develop.
Increased demographic heterogeneity in organizations has been expected to generate important
benefits, such as increasing the variance in perspectives and approaches to work brought by
members of different identity groups. Given the
purported advantages, managers might eagerly
incorporate workforce diversity into organizational problem-solving processes. Yet attempts to
capitalize on these advantages have met with
mixed success (e.g., Heilman, 1994). Likewise,
research on the effects of demographic heterogeneity in organizational settings has been characterized by mixed findings, leading researchers
to conclude that, in spite of the recent popularity
of demographic heterogeneity as a topic, there is
little consensus about either what constitutes diversity or how it affects performance (Guzzo &
Dickson, 1996: 331).
We address these inconsistencies by suggesting
that past researchers have neglected to consider
whether demographic heterogeneity among work
group members led to the emergence of certain
norms that subsequently influenced work processes and outcomes. Drawing on self-categorization theory, we begin by exploring how demographic heterogeneity influences the emergence
and stability of a group's emphasis on cooperative
norms. We then consider the relative impacts of
THEORY AND HYPOTHESES
Group Norms: The Relative Emphasis on
Cooperation
Group norms, defined as legitimate, socially
shared standards against which the appropriateness of behavior can be evaluated (Birenbaum &
Sagarin, 1976), influence how a group's members
perceive and interact with one another, approach
decisions, and solve problems. Norms are regular
behavior patterns that are relatively stable and expected by group members (Bettenhausen & Murnighan, 1991: 21). Cooperative group norms, in particular, reflect the degree of importance people
place on their personal interests and shared pursuits (Wagner, 1995: 153), shared objectives, mutual interests, and commonalties among members.
Emphasizing independence, rather than cooperation, causes people to differentiate themselves from
others and focus on their own and others' unique
interests, abilities, and characteristics.
Group cooperation may be dictated by the characteristics of a group's task but, more typically, a
work team's objectives are specified at its incep-
We thank Charles O'Reilly, Anne Tsui, Tom Lee, and
our AMJ reviewers for constructive comments on this
article. We thank the Citigroup Behavioral Science Research Council and the Institute of Industrial Relations at
the University of California, Berkeley, for their financial
support.
956
2001
Chatmanand Flynn
tion, and the means of accomplishing those objectives are left to the team's discretion (Hackman,
1987). Further, outcome interdependence is distinct from task interdependence (Wageman, 1995).
Thus, even when incentives and rewards are allocated to a group, variations are likely to emerge in
the level of interdependence members exhibit in
accomplishing the task.
A Self-Categorization Approach to
Understanding the Effects of Demographic
Composition on the Emergence and Stability
of Cooperative Norms
Much is known about the behavioral consequences of cooperative orientations, but researchers know relatively little about the factors that influence the emergence of cooperative norms. Given
the impact of cooperative orientations on processes
and outcomes in organizations and the importance
of matching a group's orientation to its task (e.g.,
Ancona & Caldwell, 1992), understanding the
emergence and stability of such norms over time is
critical. Examining group composition at the time
groups form and how members categorize themselves and other members on the basis of their
demographic differences may shed light on variations in cooperative orientations in different groups
at different times.
Self-categorization is the process by which people define their self-concepts in terms of membership in social groups. Self-concepts are activated
and provoke specific behaviors depending on the
characteristics of the others who are present in a
situation (e.g., Markus & Cross, 1990). People often
use immediately apparent physical features, such
as race, sex, and national origin, to categorize others and predict their behavior. Further, members of
demographically heterogeneous groups are more
likely to categorize one another in terms of demographic characteristics than are members of homogeneous groups (Stroessner, 1996).
The principle of functional antagonism describes
an inverse relationship between the salience of different social categories: as one category becomes
more salient, others become less salient (e.g.,
Turner, Oakes, Haslam, & McGarty, 1994). This
principle implies that when demography is salient,
a group of people will focus more on their differences than on their similarities; that is, they will be
less likely to acknowledge and act in accordance
with factors that tie them together. Research has
shown that demographic heterogeneity within
work groups is inversely related to members' focus
on organizational objectives (Chatman, Polzer, Barsade, & Neale, 1998). We suggest that this focus on
957
differences will lead to the formation of norms that
highlight individual members' interests; such
norms would be independent, rather than interdependent, norms.
One of the few studies that focused on norm
formation in work groups showed that their norms
formed early, often before groups' members adequately understood their tasks (Bettenhausen &
Murnighan, 1985). Norms were subject to modification over time, however. As group members interacted, shared experiences formed the basis for
norms governing future interactions. Interestingly,
demographic heterogeneity may influence the stability of cooperative norms as "social targets initially activate primary or primitive generic categories such as race, gender, and age" (Messick &
Mackie, 1989: 54; emphasis added). A negative relationship between demographic heterogeneity and
cooperative norms may be most pronounced early
in a group member's tenure or in a group's formation. It is during this early period that people will
have the fewest alternative, potentially competing,
categories on which to focus, given a lack of prior
knowledge about their colleagues (Brewer & Miller,
1984). The negative influence of demographic differences on group members may weaken over time
as other social categories, which were not immediately apparent, surface (Harrison, Price, & Bell,
1998).
The dominant paradigm in demography research, similarity-attraction theory, cannot account
for such temporal changes in the demographybehavior relationship. The similarity-attraction
model implies that a stable relationship exists between demographic characteristics and behavior
because similarity remains constant (e.g., Pfeffer,
1983). Social categorization theory, in contrast,
provides a more dynamic explanation in which it is
recognized that attention to specific characteristics
in a given situation may change over time. At first,
demographically different team members may be
hesitant to cooperate with one another because
they categorize each other as out-group members.
However, if the salience of surface-level demographic characteristics dissipates over time and demographically dissimilar group members begin to
recategorize themselves as fellow in-group members, they may be more inclined to cooperate with
one another. Drawing on social categorization theory and invoking the functional antagonism principle, we therefore predict the following:1
1 Predictions at both the individual and group levels of
analysis are relevant because a single member's perceptions of group norms may depend on the extent to which
958
Academy of ManagementJournal
Hypothesis la. The negative relationship between being demographically different from
the other members of a group and perceptions
that group norms emphasize cooperation will
be strongest for members who are new to the
group.
Hypothesis lb. Group heterogeneity will be
most negatively related to cooperative norms
early in a group's existence.
Past researchers have found that recategorization
was facilitated by increased contact among group
members (Sherif, Harvey, White, Hood, & Sherif,
1954). Simple contact between people with different backgrounds, here based on demographics, may
not, however, be enough to reduce biases or increase trust. To induce group members' recategorization of different people into a common in-group
identity, the contact situation must reflect certain
conditions, including, most importantly, an objective that makes members' shared fate salient
(Dovidio, Gaertner, & Validzic, 1998). This should
influence members to perceive themselves as one
superordinate group rather than as individuals differentiated by demographic characteristics. Interaction under such conditions of shared fate can
broaden perceptual fields to allow impressions of
out-group members to become more accurate and
favorable (Islam & Hewstone, 1993).
Research on the contact hypothesis has focused
on identifying the antecedent conditions that are
necessary for recategorization to occur (e.g., Brewer
& Brown, 1998). In much of this work, it has been
assumed that all group members react similarly to
increased contact. We argue, instead, that increased
contact among similar and dissimilar members may
differentially affect cooperative norms. That is,
more contact among heterogeneous team members
may lead to a higher increase in cooperative norms,
but contact may have a less significant effect among
homogeneous members.
The psychology underlying this prediction can
be illustrated by considering a high level of demographic dissimilarity at the individual level. Imagine a five-member group with one black African
woman and four white American men. She is demographically different from everyone else, and
they are demographically homogeneous with each
other. We predict that more frequent contact will
he or she agrees with prevailing group norms, and the
impact on group activities and outcomes of norms favoring independence or cooperation could depend on
whether group members unanimously or partially share
norms (e.g., Kenny & LaVoie, 1985).
October
improve relationships among the people in the
group who differ from each other-the people who
know less about one another. We assume that negative stereotypes about demographically different
people are generally untrue and that demographically similar people will respond reasonably to disconfirming information about negative stereotypes
(e.g., Stephan & Brigham, 1985). Thus, the black
African woman will, with increasing opportunities
to learn about them, improve her relationship with
all four white American men following contact,
whereas each demographically homogeneous team
member will improve only one relationship. In this
way, the perception of team norms for those who
are more demographically similar is not as affected
by increased contact as it is for those who are more
demographically different.
Similarly, at the group level, increased contact
will have less impact on demographically homogeneous groups because such interaction will merely
confirm members' positive impressions of similar
others and strengthen group cohesion. Conversely,
contact will have a greater impact on diverse
groups because members' increased contact will
provide information to members that disconfirms
unfounded stereotypes, as we predict in Hypotheses 2a and 2b:
Hypothesis 2a. With their initial (early membership) perceptions of cooperative group
norms controlled for, the perceptions of cooperative norms of demographically heterogeneous group members will increase more as a
result of contact than will the perceptions of
homogeneous members.
Hypothesis 2b. With initial cooperative group
norms controlled for, cooperative norms will
increase more as a result of contact in heterogeneous than in homogeneous groups.
The Role of Cooperative Norms in Mediating
the Relationship between Diversity and Group
Processes and Outcomes
Research has found that demographic heterogeneity influences work processes and outcomes, but
it is unclear whether it promotes or constrains
group effectiveness. On the one hand, a "value in
diversity" hypothesis has been supported. Compared to homogeneous groups, members of demographically heterogeneous groups behaved more
cooperatively (Cox, Lobel, & McLeod, 1991), were
more innovative (O'Reilly, Williams, & Barsade,
1997), and derived higher-quality solutions (e.g.,
Kirchmeyer & Cohen, 1992). On the other hand,
Williams and O'Reilly (1998) concluded that in-
2001
Chatman and Flynn
creased diversity typically has negative effects on
individual and group behavior, citing studies
showing that homogeneous, not heterogeneous,
groups were more cooperative (e.g., Alagna, Reddy,
& Collins, 1982), were more innovative (e.g.,
O'Reilly & Flatt, 1989), performed better (e.g.,
Watson, Kumar, & Michaelsen, 1993), and experienced less turnover, alienation, and dissatisfaction
among members (e.g., Tsui, Egan, &O'Reilly, 1992).
One reason for these diametrically opposed results may be that researchers have often neglected
to specify the psychological mechanisms underlying the relationship between demographic heterogeneity and work processes and outcomes, relying
instead on demographic characteristics as proxies
for such mechanisms (e.g., Lawrence, 1997). A
growing number of studies have acknowledged the
complexity of demographic effects and identified
factors that influence whether demographic diversity enhances or detracts from performance,including group structure, process, conflict (e.g., Jehn,
Northcraft,& Neale, 1999), and organizational culture (Chatmanet al., 1998). Interestingly, in many
of these studies, particularlythose relating to cooperation and conflict, norms are central to the group
composition-outcome link. For example, Simons
(1995) found that demographic heterogeneity was
only advantageous when teams were able to manage conflict. Members' tolerance for others' points
of view influenced the link between compositional
heterogeneity and creative solutions (Hoffman,
Harburg,&Maier, 1962). And cooperative cultures
in which members' shared fate was salient determined whether group heterogeneity influenced
work effectiveness (e.g., Chatman et al., 1998).
This discussion suggests that, in addition to being influenced by composition, a group's emphasis
on cooperative norms may mediate the relationship
between demographic heterogeneity and work processes and performance. Although heterogeneity
influences cooperative norms, it may be less potent
than norms in affecting work processes and outcomes. As we proposed above, salient demographic
differences influence members' behavior early in a
group's existence. This influence may occur because of a lack of knowledge of group norms
(among new members) or because of an absence of
specific group norms (in newly formed groups)
(e.g., Levine & Moreland, 1991). But, as a group
forms more specific norms, these norms may
eclipse the use of demographic differences as proxies for how members should act and treat others. As
time passes, group norms strengthen, and the degree to which they are enforced intensifies, while
members' tendencies to react to surface-level differences dissipate (e.g., Harrison et al., 1998).
959
Below we discuss how group effectiveness may
be more influenced by cooperative norms than by
the composition of members' demographic characteristics. Our focus in this section, therefore, shifts
from the early period of norm formation to the
cumulative impact of cooperative norms, because
all aspects of a team's history contribute in some
way to its processes and outcomes (e.g., Gersick,
1988).
Project and task timing. Communication frequency and timing influence group effectiveness.
Gersick (1988) found that successful teams engaged
in the most concentrated debate and discussion at
the midpoint of their life spans. Researchers have
linked demographic heterogeneity to team members' willingness to communicate about task information, assuming that demographic attributes
provide surrogatemeasures for the common experiences and backgrounds shaping communication
(Pfeffer,1983) and that people may be less inclined
to share task information with those who are demographically different (e.g., Zenger & Lawrence,
1989). In contrast, we propose that communication
frequency and timing may be more influenced by
the norms a team adopts than by its demographic
heterogeneity.
Regardless of demographic heterogeneity, more
interaction occurred among members in groups
that were experimentally manipulated to emphasize norms supporting interdependence ratherthan
independence (Chatman et al., 1998). Driven by
salient group objectives and greater agreement
about how to approach required tasks, members of
groups that emphasize cooperative norms may be
more likely to meet as soon as their tasks are assigned, giving them more time to consider and determine group processes. In contrast, teams that
develop less cooperative norms may fail to recognize the need to address procedural issues that go
beyond dividing up the tasks. Members of such a
team may, instead, assume that the group is merely
a collection of individuals who will each work on a
part of the task independently. Such an approach
may hinder the effective accomplishment of interdependent tasks, but members may not realize the
need for cooperation until project deadlines are
close and members reconvene to integrate the distinct pieces of their projects. As a result, groups
emphasizing less cooperative norms will be more
likely to experience a flurry of group meetings at
the point when the tasks need to be completed than
will teams emphasizing more cooperative norms.
We predict that cooperative norms will mediate the relationship between demographic heterogeneity and the timing of group communication
and interaction. Specifically, regardless of demo-
960
Academy of ManagementJournal
graphic composition, teams emphasizing cooperative norms will be more likely to hold early
meetings (those occurring closer to the time that
the tasks are assigned) than late meetings (those
occurring closer to the time that the tasks must be
completed) than will teams not emphasizing
cooperative norms, as stated in Hypotheses 3a
and 3b:
Hypothesis 3a. Perceptions of cooperative
norms will be positively related to a group's tendency to hold more early than late meetings.
Hypothesis 3b. Perceptions of cooperative
norms will mediate the relationship between
demography and early versus late meetings in
such a way that the direct effect of group heterogeneity will weaken after cooperative norms
are considered.
Members' satisfaction and intentions to remain
with a team. Norms supporting cooperation may
also have a greaterinfluence than demographicheterogeneity on team members' attitudes about a
team experience. Research has shown that people
who are more demographically different from their
coworkers are less satisfied and less likely to intend
to remain in their jobs (e.g., O'Reilly, Caldwell, &
Barnett, 1989). However, Tsui and her colleagues
(1992) found that some of the negative effects of
being demographically different disappeared when
they took "company effects" into account. They
concluded that corporate culture or other firmspecific characteristics contributed to satisfaction
and reversed the relational demography effects.
This reversal may occur because members of teams
that emphasize cooperative norms may be more
satisfied with their team experience because of the
team's emphasis on satisfying members' needs and
objectives and maintaining harmonious relationships. Conversely, the more team members believe
that they can perform effectively as independent
contributors, the less likely they are to value team
membership, regardless of the demographicheterogeneity of the group. Thus, less conflict regarding
members' contributions may arise in teams emphasizing cooperative norms. Such positive team interaction will likely lead to members of more cooperative teams being more satisfied than members of
teams that develop less cooperative norms, as we
predict in Hypotheses 4a and 4b:
Hypothesis 4a. Perceptions of cooperative
norms will be positively related to group members' satisfaction with group processes and
products.
October
Hypothesis 4b. Perceptions of cooperative
norms will mediate the relationship between
members' demographic heterogeneity and satisfaction in such a way that the direct effect of
heterogeneity on satisfaction will weaken after
cooperative norms have been considered.
Individual performance and team efficiency
and effectiveness. Past research results have been
characterized by inconsistent links between demographic differences and work effectiveness. This
discrepancy may be reconciled by considering the
mediating role of cooperative norms. A lack of emphasis on interdependence among members may
render uncooperative teams less efficient and less
effective in accomplishing their objectives than
teams that emphasize cooperative norms. Because
group-level goals are emphasized in cooperative
teams, members are likely to develop shared views
about ways to approach and accomplish their required tasks. In contrast, less cooperative groups
may be forced to spend more time synthesizing
team members' perspectives because task accomplishment proceeds more efficiently and effectively
when team members agree about the proper way to
approach a task (Bettenhausen &Murnighan, 1985).
Ancona and Caldwell (1992) found that greater
tenure heterogeneity impeded innovation among
product development teams by reducing their capability for teamwork and, subsequently, their ability to get new products implemented quickly. This
observation suggests that demographic heterogeneity influences work group efficiency and effectiveness more indirectly, through an influence on cooperative norms. In contrast, norms may influence
individual and team output, such as quality, productivity, and creativity, directly (e.g., Chatman et
al., 1998), even if members have the requisite variety of ideas to achieve high levels of success in
addressing complex tasks. When less cooperative
norms prevail and a team is characterized by a
focus on individual achievement, self-interest, and
a lack of commitment and trust, members may be
inhibited about sharing their ideas (Gaertner,
Mann, Dovidio, Murrell, & Domare, 1990). In teams
that are less cooperative, sharing information and
ideas is risky because of the potential for the dilution of the individual credit and rewards granted
for those ideas. Conversely, members of teams emphasizing cooperative norms will be more willing
to contribute their ideas because their rewards are
derived from meeting team goals. At the group
level, a cooperative emphasis will improve performance by focusing members' efforts on a single
objective. A less cooperative team, in contrast, may
encounter difficulty integrating individual contri-
Chatmanand Flynn
2001
butions into a cohesive final product. Thus, regardless of demographic heterogeneity, team members'
enhanced focus on group-level goals may yield
greater team efficiency and effectiveness and individual performance,with such norms eclipsing the
direct influence of heterogeneity on these outcomes, as we predict in Hypotheses 5a-6b:
Hypothesis 5a. Perceptions of cooperative
norms will be positively related to individual
contributions to group objectives.
Hypothesis 5b. Perceptions of cooperative
norms will mediate the relationship between
demographic heterogeneity and members' contributions to group objectives in such a way
that the direct effect of heterogeneity will
weaken after perceptions of cooperative norms
are considered.
Hypothesis 6a. Cooperative norms will be positively related to team efficiency and effectiveness.
Hypothesis 6b. Cooperative norms will mediate
the relationship between demographic heterogeneity and team efficiency and effectiveness
in such a way that the direct effect of heterogeneity will weaken after cooperative norms
are considered.
STUDY 1 METHODS
Respondents and Study Design
One hundred nineteen students, representing
half of the first-year class enrolled in a two-year
full-time M.B.A. programat a majorAmerican university, participated in this study. All students who
were asked to participate in the study did so, yielding a 100 percent response rate. Thirty-six percent
were not U.S. citizens, 22 percent were nonwhite,
and 32 percent were women. Respondents' mean
age was 28.60 years, and they had an average of
5.31 years of full-time work experience. The demographic profile of study participants was identical
to that of the entire first-year class.
As part of their required organizational behavior
course, students had to complete a semester-long
consulting project in five-person groups, which,
with an associated presentation, ratings by other
team members, and a paper describing the team
process, accounted for 42.5 percent of their final
course grades. Students assembled their own
groups during the first two weeks of classes (n = 24
groups). For the remainder of the 15-week semester, each group identified and addressed a critical
organizational behavior problem confronting a real
961
organization of their choosing. At the end of the
semester, each team was required to submit a written report of its consulting analysis and recommendations and present an oral report of findings to the
other students in the class and the instructor;many
also had members of the firms on which their
projects focused present.
Once the project of a student's group was completed, but before graded projects were returned to
the groups, each student submitted a four-page paper describing his or her team experience. Students
were informed that their evaluation papers would
be assessed in terms of their ability to constructively evaluate their team's work approach. These
papers were content-coded for the present study by
two independent judges who were blind to the
hypotheses. Specific procedures are described in
the dependent variables section below.
Survey data regardingthe team experience were
collected at two points in time. Respondents assessed teams' cultural norms during the 3rd (time
1) and 15th (time 2) weeks of the 15-week semester,
with each independently reporting perceptions of
her or his team's current relative emphasis on cooperative norms. Response rates were 98 percent at
time 1 and 100 percent at time 2. These data were
collected in class, and no advance warning was
given, to avoid cueing effects. Respondents were
told that these data would be used to "assess the
effectiveness of using groups in the M.B.A. core
curriculum" and that course grades would in no
way be affected by these data, their instructor
would not see their responses, participation was
voluntary, and responses would remain completely
confidential. Assessments of each team's interaction during the semester were collected at time 2.
Each team member was required to provide independent ratings of himself or herself and of each
other team member on a variety of dimensions on
the last day of the semester. Finally, demographic
data were collected from the school's archives.
Independent Variables
Demographic heterogeneity. At the individual
level, we calculated relational demography scores
to reflect sex, race, and citizenship differences between individuals and other team members. Following others (e.g., Tsui et al., 1992), we used this
Euclidian distance measure: [1/nl(xi - j)2]1/2,
where xi equaled the focal individual's score on a
dimension (0 = "male,"1 = "female";0 = "white,"
1 = "nonwhite"; and 0 = "U.S. citizen," 1 = "not
U.S. citizen");xj equaled each other team member's
score on that dimension; and n equaled the number
of students in the focal student's group. A rela-
962
Academy of Management Journal
tional measure ranging from 0 to 1 was derived for
each of the three demographic dimensions. Social
categorization theory focused us on visible differences in themselves rather than on their content,
which has been the focus in prior research (e.g.,
Riordan & Shore, 1997). In this sense, individual
demographic differences are best interpreted, at
least in relational demography terms, as an amalgamation (e.g., Wayne & Liden, 1995). Therefore,
we summed the three individual relational scores
to create an overall measure of relational demography. Calculated scores ranged from 0.0 (similar to
others) to 2.44 (different from others). The higher
the overall relational demography score, the more
demographically different respondents were from
other people within their teams in terms of citizenship, race, and sex (x = 1.65, s.d. = 0.46).
At the group level, we used the coefficient of
variation, a scale-invariant measure of dispersion
(Allison, 1978), to assess the relative demographic
heterogeneity of the team. We calculated the coefficient of variation for the M.B.A. groups as the
standard deviation of the total demographic difference between team members divided by the mean
of the total demographic difference between team
members (x = 0.79, s.d. = 0.43). Groups ranged on
this measure from a score of 0.0 (homogeneous) to
a score of 1.48 (heterogeneous).
Contact among team members. At time 2, respondents reported the frequency of their teams'
interactions over the semester in terms of the numbers of meetings held. Respondents reported the
number of team meetings that occurred for three
periods: the 1st through the 8th week of the semester, a meaningful end point, because during their
8th week all M.B.A. students participated in an
intersession programfor which they were excused
from their regular classes (x = 3.48, s.d. = 1.79); the
gth through the 14th week (x = 6.06, s.d. = 3.11);
and the 15th week of the semester (x = 3.05, s.d. =
1.76). We created a measure of the total number of
meetings each team held by summing these three
measures at the individual level (x = 12.66, s.d. =
4.64) and averaging team members' scores at the
group level (x
=
12.68, s.d. = 3.17).2
We calculated the corresponding eta-square for all individual variables aggregated to the group level to check
whether any two people within the same group responded
more similarly than two people in different groups. The
eta-square far exceeded Georgopoulos's (1986: 40) minimum criterion of .20, supporting aggregation.
2
October
Dependent Variables
Cooperative team norms. Perceptions of teams'
norms were assessed twice, as described above.
Each respondent independently reported how
strongly each of five statements characterizedhis or
her team using seven-point Likert-type scales (1,
"strongly disagree,"to 7, "strongly agree").The cooperative norms scale contained the following
items (tense was present for time 1 and past for
time 2,;the 15th and last week of the semester): (1)
"It is/was important for us to maintain harmony
within the team," (2) "There is/was little collaboration among team members, tasks are/were individually delineated" (reverse-coded), (3) "Thereis/
was a high level of cooperation between team
members," (4) "People are/were willing to sacrifice
their self-interest for the benefit of the team," and
(5) "There is/was a high level of sharing between
team members." The overall reliability (alpha) coefficients for the scale were .62 at time 1 and .77 at
time 2. Responses to the items summed for each
time period yielded an overall measure of each
respondent's perception of his or her team's norms
pertaining to cooperation. This individual-level
measure is thereafter referred to as perceptions of
cooperative norms. Perceptions of cooperative
norms at time 1 (x = 24.15, s.d. = 3.62) and time 2
(x = 24.60, s.d. = 4.53) were used as dependent
variables to test Hypotheses la and 2a. We averaged
norms at time 1 and time 2 (x = 24.47, s.d. = 2.98)
to test the individual-level main and mediation
hypotheses (Hypotheses 4a and 4b and 5a and 5b).
We developed a measure of norms at the group
level by calculating the average, within groups, of
all members' scores on the cooperative norm scale
for each time period. This group-level measure is
hereafterreferredto as cooperative norms. Cooperative norms at time 1 (x = 24.61, s.d. = 2.99) and
time 2 (x = 24.19, s.d. = 3.46) were used as depen-
dent variables to test Hypotheses lb and 2b. We
averaged each team's cooperative norms over the
two time periods (x = 24.47, s.d. = 2.98) and used
this measure as an independent variable to test the
group-level main and mediation hypotheses (3a,
3b, 6a, and 6b).
Earlyllate meetings. To assess whether groups
held more meetings early or late in the 15-week
period, we created a ratio with the average number of meetings held each week during the first 8
weeks as the numerator and the average number
of meetings held during the 15th week as the
denominator
(x = 0.19, s.d. = 0.14). A larger
value indicated a team met more often earlier
than it did later.
2001
963
Chatman and Flynn
Satisfaction, performance, and team efficiency
and effectiveness. Satisfaction, efficiency, and effectiveness were assessed through content coding
the team evaluation papers mentioned above. In
this assignment, students were told to "evaluatethe
effectiveness of your teams' approach to the consulting project."To reduce social desirability bias,
we informed students that their assessment would
be based on their analyses, not on how effective the
teams were or how satisfied the students were with
them. They were instructed verbally and in writing
to comment on: (1) team effectiveness, (2) whether
certain biases or conflicts hindered team efficiency,
and (3) how satisfied they were with their experience and final results. Coders were instructed to
look for these categories of comments in the papers
and were easily able to do so, as the high interrater
reliabilities reported below indicate.
Forthe content analysis, we removedall references
to team members' names, sex, race, and nationality
from the evaluation papers. To retain continuity, we
assigned each member a two-digit identification
number,which was used in place of personal identifiers. Two independent coders,who were blind to the
study hypotheses, then read the papers (averagein-
averaged teammates' ratings of each focal individual (excluding self-ratings) to obtain a single composite measure of each person's overall contribution to the team project (x = 7.54, s.d. = 1.09). We
used a single-item within-group interraterreliability equation (rWG(1)= 1 - [sXj//EU2I; James, Demaree, & Wolf, 1984) to assess the reliability of these
individual contribution ratings. A reliability score
was obtained for each group of raters (n = 119; x =
0.79, s.d. = 0.14 for the reliability scores).
Coders rated team effectiveness (defined as "the
extent to which the team accomplished its purpose
and produced the intended, expected, or desired
result") and team efficiency ("the extent to which
the team functioned with the least waste of time
and effort")(1, "highly inefficient or ineffective," to
terrater agreement across the four variables = .86).
Control variables. At the individual level of
analysis, dichotomous dummy variables for citizenship, race, and sex were included in each re-
Coders were provided with definitions of each construct, and disagreementsover the interpretationof
these terms were resolved by discussion between
coders after they had completed a set of practice
papers (the same assignment had been collected in
the course the prior year). They coded all papers
within a single team before moving on to the next
team, rating individual-level variables immediately
after reading each evaluation paper and group-level
variablesafterreading all five randomly orderedpapers from membersof a single team. The exact measures are described in detail below.
Two measures of satisfaction (defined for coders
as "the collection of both positive and negative
feelings and beliefs that people have about their
team experience") were assessed (1, "very dissatisfied," to 7, "very satisfied"). Coders rated each respondent's satisfaction with the team process (x =
4.92, s.d. = 1.35) and team product (x = 5.73, s.d. =
1.09). These measures, which were highly correlated (r = .79, p < .01), were then averaged into an
overall measure of individual satisfaction (x
=
5.32, s.d. = 1.15).
Individual performance was assessed with rat-
ings of each team member made by his or her teammates. Each member was asked to report the extent
to which a focal individual "contributed to the
successful running of and outcomes generated by
the team" (1, "made minor or no contributions," to
9, "made major, extensive contributions").We then
7, "highly efficient or effective"; Xteamefficiency
5.04,
s.d. = 1.44; Xteameffectiveness = 4.29,
=
s.d. =
1.81). Because the two measures were highly correlated (r = .90, p < .01), we averaged them to
create an overall measure, team effectiveness and
efficiency
(x = 4.67, s.d. = 1.58).
Model
gression equation. (0 = "U.S."/"white"/"male"; 1 =
"non-U.S."/"nonwhite"/"female").Doing this controlled for the possibility that demographic patterns affected one nationality, race, or sex more
than others, consistent with our focus on the effects
of being demographically different, rather than on
the content of such differences. We controlled for
team size at the group level of analysis because
slight variations existed (three groups of four and
two groups of six members). In tests of Hypothesis
2a, we controlled for perceptions of group norms
and group norms at time 1 to determine whether a
regression to the mean influenced changes in demographically different people's perceptions and
heterogeneous teams' norms, which, in Hypothesis
la, were predicted to be more extreme (negative).
Analyses. We used hierarchical regression to test
the hypotheses at the individual and group levels
of analysis. Control variables were entered in the
first step, demographic heterogeneity (relational
demography for the individual level and the coefficient of variation for the group level) in the second step, and individual- or group-level cooperative norms in the third step. We used slope
analyses (e.g., Schoonhoven, 1981) to test the mod-
erating hypotheses (2a, 2b), entering contact in the
third step and the interaction of demographic heterogeneity and contact in the final step. To test the
964
Academy of ManagementJournal
mediation hypotheses (3a-6b), we used a series of
three regression analyses: (1) the dependent variable regressed on the independent variable, (2) the
mediator regressed on the independent variable,
and (3) the dependent variable regressed on the
independent variable simultaneously with the mediator variable. Mediation is demonstrated to the
extent that the mediator (cooperative norms) relates
to the dependent variable (timing, satisfaction, effectiveness) beyond the effect of the independent
variable (demographic heterogeneity). If a variable
is a mediator, the level of significance for the coefficient of the independent variable will decrease,
and the significance of the mediator variable will
remain constant when the two variables are entered
simultaneously (Baron & Kenny, 1986).
STUDY 1 RESULTS
Means, standard deviations, and correlations are
reported in Table 1.
Predicting Cooperative Group Norms
Hypothesis la predicts that the impact of demographic heterogeneity on perceptions of cooperative norms will be strongest when members are
new to a group and will decrease over time. We first
found that each separate demography measure had
a significant, negative effect on perceptions of cooperative norms in the M.B.A. sample (I3citizenship =
-0.19, p < .05, gBrace = -0.19, p < .05; I3sex
-0.15, p < .10), empirically justifying aggregation.
Table 2 shows the results of tests of this hypothesis
and of the other predictions made in the first two
sets of hypotheses. Relational demography was significantly, negatively related to perceptions of cooperative norms at time 1 (,3 = -0.35, p c .01,
equation 1) but not at time 2 (,3 = -0.07; n.s.,
equation 2). To further test whether the impact of
demographic heterogeneity on perceptions of cooperative norms was significantly less at time 2 than
at time 1, we pooled responses for both times and
consolidated measures of cooperative norms at
both times into a single variable. We then ran a
hierarchical regression equation in which simple
demographic controls and a dummy for time ("time
1" = 0, "time 2" = 1) were entered in the first step,
demographic heterogeneity in the second step, and
the interaction of demography and time in the third
and final step. The coefficient of the interaction
term (,B = 0.34) was marginally significant (p <
.10), indicating that the influence of relational
demography on norms was greater at time 1 than
time 2.
Hypothesis ib, stating that the impact of team-level
October
demographic heterogeneity on cooperative norms
will decrease over time, was supported, since team
heterogeneity negatively and significantly influenced
cooperative norms at time 1 (,3 = -0.45, p ' .05,
equation 3) but not at time 2 (f3 = -0.17, n.s., equation 4). Using the test for Hypothesis la described
above, we found the effect of team heterogeneity on
cooperative norms was marginally, significantly less
at time 2 than time 1 (,3 = 0.75, p ' .10).
Predicting Changes in Team Norms as a Function
of Intragroup Contact
Hypothesis 2a predicts that demographically different people will be more affected by contact with
team members in the extent to which perceptions
of cooperative norms change (increase) over the
course of a project. As predicted, equation 2 in
Table 2 shows a significant, positive coefficient for
the interaction of amount of contact and relational
demography at time 2 (,3 = 1.17, p c .05). To
understand the form of this interaction, we used
the slope analysis equation (Schoonhoven, 1981):
y = b1x1 = b3x1x2, where b1 is the unstandardized
coefficient for relational demography, x1 is the relational demography score, b3 is the unstandardized coefficient of the interaction of contact and
demography, and x2 is the level of contact. We
calculated the contribution of demographic heterogeneity at two levels of contact: relatively low for
this sample (one standard deviation below the
mean, or 8.02) and relatively high (one standard
deviation above the mean, or 17.30). For the interaction between level of contact and relational demography, an increase in relational demography
from one standard deviation below the mean (1.21)
to one standard deviation above the mean (2.09)
increased the perception of cooperative norms by
4.41 more when a respondent reported a higher
level of contact among team members, as revealed
in the detailed calculations: For less contact,
-.67(1.21) + .54(1.21)(8.02) = 4.43, and -.67
(2.09) + .54(2.09)(8.02) = 7.65. For more contact,
= 10.49, and
+ .54(1.21)(17.30)
-.67(1.21)
-.67(2.09) + .54(2.09)(17.30) = 18.12. The important number is the difference of the differences at
time 2 ([18.12 - 10.49] - [7.65 - 4.43] = 4.41),
representing the increase in cooperative norms for
a two-standard-deviation increase in relational demography for those experiencing more rather than
less contact.
Hypothesis 2b, which predicts that demographically heterogeneous groups will be characterized by a
greater change (increase) in cooperative norms over
time when members have more contact but that homogeneous groups will evidence less variance in
a
** *
9. 8. 7. 6. 5. 4.
p p bn=
Group
< <
Total
n
116.
.01 .05
=
24;
3.
2.
1. Study14.13.12.11.10.9. 8. 7. 6. 5. 4.
2.
2b
1. Study
2'
Sex
Race
Total Team Team
Team Total
U.S.
U.S.
Male
Male
White
White
Early/late
Female
Relational
sizeFemale
Relational
Individual
Individual
Citizenship
Citizenship SatisfactionCooperative
Perceptions
Satisfaction
Cooperative
NonwhiteNon-U.S.
NonwhiteNon-U.S.
of
meetings,
meetings,
Total
Sex
Race
demography
performance
individual
n
cooperative
increase
119.
Variable
composition
effectiveness
meetingsnorms,
norms,
group
demography
and
performance
group
individual
compensation compensation
=
3.
individual
norms
efficiency
1.21
2.32
4.64
23.66
0.18
12.40%
9.40%
22.90%
77.10%
87.60%
91.60%
$226,948
5.32
4.96
24.37
7.544.65
12.69
12.66
24.40
0.791.65
0.19
36.10%
31.90%
21.80%
63.90%
78.20%
68.10%
Mean
Means,
s.d.
1.150.14
0.46
4.64
3.16
2.303.170.430.46
1.091.58
0.51
7.73
0.37
0.50
1.50
$144,682
Standard
-.06-.05-.04-.14 .03
.38**
-.05
-.01-.02-.02
-.18* -.06
.42**
.09
-.17.11.02.13.03.04.10.17.10.11.05
-.07
-.08.13.00.14.12-.12.08.03-.35 -.18
.48**
.21*
.15**
-.19*
1
2
Deviations,
-.06-.02.06-.02 -.09
.25**
-.12-.16-.10-.18.14-.01-.10.00-.15 .07
.40**
3
-.01
.16*
-.19**
.29**
-.21**
-.11-.36-.17-.19.14-.05-.15-.12.16.22
4
-.09-.08.03
-.13-.13 -.39.41*.18-.20-.14.02
-.18*
5
.14.17*
.32**
.04-.32-.22-.40-.25.14-.43*
.41*
6
.07.06
.04-.03
.20*
.73**
.59**
.56**
.58**
7
.03
.36
-.03-.33
.63**
.55**
.81**
8
.21**
.06 -.05-.04-.27
.68**
.14 .14-.10-.16
.27 .50*-.05
.26**
.65**
.51*
and
TABLE
1
Correlations
among
9
10
11
12
13
14
Study
Variables
966
Academy of Management Journal
October
TABLE 2
Results of Hierarchical Regression Equations Predicting Perceptions of Cooperative Norms and
Cooperative Norms, Study la
Perceptions of Cooperative
Norms, Hypotheses la and 2a
Variable
Time 1:
Equation 1
Citizenship
Race
Sex
Cooperative norms, time 1
Team size
AR2
0.17*
0.03
-0.04
Relational demography
Team composition
AR2
-0.35**
0.03
0.08
Total meetings
AR2
Relational demography X total meetings
Team composition X total meetings
AR2
R2
Overall F
df
Time 2:
Equation 2
Cooperative Norms, Hypotheses
lb and 2b
Time 1:
Equation 3
Time 2:
Equation 4
-0.16
0.03
0.00
-0.05
0.00
-0.45*
0.19
-0.17
0.03
0.09
0.06
0.04
0.19*
0.05
-0.07
0.01
0.08
0.03
0.00
0.00
1.17*
-0.75
0.04
.11
3.40
4, 113
.10
1.69
7,110
0.05
.22
2.97t
2, 21
.08
0.32
5, 18
a Entries represent standardized
coefficients.
t p c .10
* p '
.05
p '
.01
**
these norms regardless of intragroup contact, was not
supported (equation 4, ,3 = -0.75, n.s.).
The Mediating Effect of Cooperative Norms on
Group Processes and Outcomes
Table 3 gives the results of regression analyses predicting early/late meetings. Hypothesis 3a was supported, because cooperative norms had a significant,
positive effect (Table 3, equation 1, ,3 = 0.56, p c .01)
and, as expected, greater team heterogeneity had a
significant, negative effect (equation 1, ,3 = -0.38,
p ' .05) on the ratio of early to late meetings. The
mediating hypothesis, 3b, was also supported, because the relationship between team heterogeneity
and early/late meetings lost significance (,3 = -0.15,
n.s.) when the cooperative norms variable was entered simultaneously, and the strength of the relationship between cooperative norms and early/late meetings remained intact. Figure 1 summarizes all the
mediating effects, which are not shown in the tables.
Since teams may have reported more cooperative
norms because they met more often, we tested
whether early norms could predict meetings during
the last week of the semester. We ran the meeting
ratio equation (equation 1) again, substituting the
meetings during the last week of the semester for
the dependent variable (ratio of early to late meetings) and our early (time 1) measure of norms for
the average across all three meeting periods. As
expected, we found a significant, negative coefficient for the cooperative norms measure (,3 =
-0.48, p < .01); thus, the more cooperative the
norms, the fewer later meetings a group had. This
finding provides some assurance that causality is in
the hypothesized direction, from norms to meetings rather than from meetings to norms.
Members' satisfaction with their team. As predicted in Hypothesis 4a, perceptions of cooperative
norms were significantly related to judges' ratings
of members' satisfaction (equation 2, ,3 = 0.59, p <
.01) and, as expected, relational demography was
negatively related to individual satisfaction (,3 =
-0.24, p c .05). Hypothesis 4b was also supported,
because the effect of relational demography on satisfaction lost significance (,3 = -0.07, n.s.) when
the measure of cooperative norms, which maintained significance, was entered simultaneously.
Individual performance and team effectiveness and efficiency. Hypothesis 5a was supported
because perceptions of cooperative norms were
positively related to peer ratings of each member's
967
Chatmanand Flynn
2001
TABLE3
Results of Hierarchical Regression Equations Predicting Team Interaction and Individual and Team
Process Outcomes and Performance, Study la
Variable
Team
Interaction,
Hypothesis 3a
Individual Process
Outcome,
Hypothesis 4a
Early/Late
Meetings,
Equation 1
Satisfaction,
Equation 2
Citizenship
Race
Sex
Team size
AR2
-0.19
.04
Relational demography
Team composition
AR2
-0.38
.14
Individual and Team Performance,
Hypotheses 5a and 6b
Peer Rating of
Individual Contribution,
Equation 3
0.02
0.02
-0.10
-0.20*
-0.09
-0.12
.01
.06
.04
.00
0.56*
.25
0.59
.32
0.25*
.06
R2
.43
5.04"
3,20
.37
12.96
5,112
.12
2.90
5,112
df
-0.36
.13
-0.03
-0.24
Cooperative norms
AR2
Overall F
Team Effectiveness
and Efficiency,
Equation 4
-0.27"
.07
0.80*
.52
.72
16.78
3,20
a
Entries represent standardized coefficients. To test the mediating hypotheses, variables from the second and third steps were entered
simultaneously in a separate set of regressions. Results from these separate regressions are reported in the text and in Figure 1.
tp c .10
*pc
.05
** p '
.01
contribution (equation 3, 1B= 0.25, p c .01). However, a test of the mediating hypothesis, 5b, was
precluded because team members' contributions
were not significantly related to how demographically different they were from their teammates.
Hypothesis 6a was supported, because teams
characterized by cooperative norms were significantly more effective and efficient (,3 = 0.80, p c
.01), as rated by independent judges. The mediating
hypothesis, 6b, was modestly supported, as demographically heterogeneous teams were marginally
less effective and efficient than were homogeneous
teams (,B= -0.27, p c .10), and this relationship
lost significance (,B= -0.06, n.s.) when the measure of cooperative norms, which maintained its
significance, was entered simultaneously.
STUDY 2, METHODS
Respondents and Study Design
We conducted study 2 to determine the extent to
which our findings fromstudy 1 could be generalized
to actual work settings. One hundred sixty-one officers from ten business units of the North American
division of a largeU.S. financial services firm participated in this study, representingan 89 percent re-
sponse rate. The participating business units conducted work in global finance. The groups ranged in
size from 9 to 58 officers (x = 28.32, s.d. = 13.91).
Respondents' mean age was 42.81 years, and average
tenure with the company was 11.24 years. Nine percent were not U.S. citizens, 12 percent were nonwhite, and 23 percent were women.
Personnel data, including compensation, performance appraisals, and demographic information,
were obtained from the company's archives. We
also collected original survey data by mailing surveys to the heads of the participating business
units, who distributed surveys to all of their officers. Respondents returned the surveys directly to
us in preaddressed envelopes and were assured
that their survey responses would be confidential
and not disclosed to their employer. Because of the
small number of business units and cross-sectional
design of this study, our analyses of these data were
limited to testing a portion of the individual-level
hypotheses (la, 4a-4b, 5a-5b).
Variables
Independent: Demographic heterogeneity. As
in study 1, differences in citizenship, race, and sex
968
Academy of Management Journal
October
FIGURE 1
Main and Mediating Effects of Demography on Individual and Team Outcomesa
(a) Individual Level
-24*/-.07
Satisfaction
.59*
~~~~-.25/-.06
/
.16
-.29* *
Relational
Demography
Perception of
/
erformance
Rating
/
Cooertive
Compensation
-_,18/-.15Increase
(b) Group Level
-.38*/-.15
.56
Team
Composition
l
-.42
Early/Late
~~Meetings
Coopeativ
Norms .0
2 7 '/_06
Team Efficiency
and Effectiveness
a Numbers above the arrowsrepresent standardizedcoefficients (betas).Betas in bold are based on regression equations including the
connectedness mediator.
.10
t p
<
.05
** p '
.01
*
p
One-tailed tests.
were assessed by comparing each individual's citizenship, race, and sex to those of every other individual in her or his business unit. Again, respondents' scores for citizenship, race, and sex
differences were combined to create an individuallevel measure of demographic differences (x =
1.21, s.d. = 0.51).
Dependent variables. Respondents rated the cooperative norms of their business units, answering
five items about the extent to which the following
described the units (1, "extremely uncharacteristic," to 9, "extremely characteristic"): cooperative,
individualistic
collectivistic,
(reverse-coded),
team-oriented, and supportive. We summed responses to form an index of cooperative norms (x =
23.66, s.d. = 7.73, a = .78).
Satisfaction was assessed through five questions
rated on a seven-point scale. Examples are "I've
thought seriously about changing organizations"
(reverse-coded) and "If I have my own way, I'll be
working for [the firm] three years from now." Responses to the items were averaged, yielding an
overall measure of each respondent's satisfaction
(x= 4.64, s.d. = 1.50, a = .81).
Individual performance was assessed with a single-item rating of overall performance (x = 2.32,
s.d. = 0.50) assigned to each officer annually by his
or her direct supervisor on a scale of 1 ("did not
meet expectations") to 3 ("exceeded expectations").
We also collected data from the firm regarding employees' total annual compensation level (salary
plus all bonuses) for the current year (x - $233,659;
s.d. = 165,672) and the year prior to the study (x =
$226,948; s.d. = 144,682). We then used prior-year
compensation to calculate compensation increases
for each individual by subtracting it from currentyear compensation and dividing that amount by
prior-year compensation (x = 0.18, s.d. = 0.37).
Control variables. Dichotomous dummy variables
for citizenship, race, and sex were included in each
equation, and we controlled for prior-year compensation in the compensation increases equation.
STUDY 2, RESULTS
Means, standard deviations, and correlations for
study 2 are shown in Table 1. Table 4 gives the
results of regression analyses for study 2.
Chatmanand Flynn
2001
969
TABLE4
Results of Hierarchical Regression Equations Predicting Perceptions of Cooperative Norms and Individual
Outcomes, Study 2a
Hypothesis
4
Hypothesis la
Hypothesis 5a
Perceptions of
Cooperative Norms
Newcomers, Equation 1
Perceptions of
Cooperative Norms,
Veterans, Equation 2
Satisfaction,
Equation 3
Individual
Performance,
Equation 4
Total Compensation
Increase,
Equation 5
Citizenship
Race
Sex
Compensation, prior year
AR2
-0.01
-0.18
-0.38k
-0.01
-0.28
-0.15
-0.03
-0.01
0.05
-0.10
-0.10
-0.01
.16
.12
.01
.02
-0.14k
-0.02
-0.07
-0.22
.07
Relational demography
AR2
-0.86
.11
0.07
.00
0.05
.00
-0.25
.02
-0.18
.01
Variable
Perceptions of cooperative norms
AR2
.27
1.78k
4, 23
Ri2
Overall F
df
.12
0.79
4, 27
0.34*
.10
0.16*
.02
0.15*
.02
.11
3.82*
5, 150
.06
1.93t
5, 150
.08
2.84"
6,149
a Entries represent standardized coefficients. To test the mediating hypotheses, variables from the second and third steps were entered
simultaneously in a separate set of regressions. Results from these separate regressions are reported in the text and Figure 1.
t p? .10
*
**
p
?
p '
.05
.01
Predicting Cooperative Norms
Hypothesis la posits that the negative relationship between being demographically different and
perceptions of cooperative group norms would be
strongest when members were new to a group. To
test this hypothesis, we created two subsamples
based on employee tenure. Those employees whose
tenure fell at least one standard deviation below the
mean were labeled "newcomer" employees (x =
1.42, s.d. = 0.56), and those employees whose tenure fell at least one standard deviation above the
mean were labeled "veteran" employees (x = 23.57,
s.d. = 4.21). Once again, each separate demography
measure was negatively related to perceptions of
cooperative norms in the financial services firm
sample (I3citizenship = -0.22, p < .05, grace = -0.07,
n.s.; 3sex = -0.26, p < .05), justifying aggregation.
For newcomer employees, relational demography
had a significant, negative effect on perceptions of
cooperative norms (equation 1, ,B= - 0.86; p ' .05).
And, for veteran employees, relational demography
had no significant impact on perceptions of cooperative norms (equation 2, ,B = 0.07, n.s.). Using an
analysis similar to that described for Hypotheses la
and lb in study 1, we again found a modestly
significant difference (,B = 0.42, p c .10) in the
demographic heterogeneity effect for the two tenure categories (0 = "newcomer," 1 = "veteran").
The Mediating Effect of Cooperative Norms on
Group Outcomes
Satisfaction. Respondents who viewed their
business units as more cooperative were more satisfied than were those who perceived the units as
being less cooperative (equation 3, ,B = 0.34, p '
.01; again, see Figure 1 for mediation results), supporting Hypothesis 4a. But demographic heterogeneity and satisfaction were not significantly related
(equation 3, ,B = 0.05, n.s.), suggesting that, rather
than playing a mediating role (and precluding a test
of Hypothesis 4b), only the direct effect of cooperative norms influenced members' job satisfaction in
this sample.
Individual performance.
Respondents who
were more demographically different from others
in their business unit were lower performers (equation 4, ,B = -0.25, p s .05; also see Figure 1) and
received less substantial compensation increases
(equation 5, ,B = -0.18, p s .05). And, supporting
Hypothesis 5a, respondents who perceived norms
to be more cooperative performed better (equation
4, ,B = 0.16, p s .05) and received higher compensation increases (equation 5, ,B= 0.15, p ' .05). We
also found a decrease in the strength of the relationship between demographic heterogeneity and
performance (,B = -0.06, n.s.) and demographic
heterogeneity and compensation increases (,B =
970
Academy of ManagementJournal
-0.15, p c .10) when the cooperative norms measure was entered simultaneously in each equation
(in both cases, the cooperative norms measure
maintained its level of significance). Given that the
significance of the demography coefficient did not
diminish completely in equation 5, we would label
this result "partial mediation" (James & Brett,
1984). Thus, we found corroborating support for
three of the four hypotheses tested in this study.
DISCUSSION
The Relationship between Demography and
Cooperative Norms
We examined whether demographic heterogeneity among team members influenced the emergence
and stability of cooperative norms and how such
norms, in turn, mediated the relationship between
demographic heterogeneity and work processes
and outcomes. Hypotheses la and lb, addressing
the individual and group levels of analysis, were
modestly supported in both samples. M.B.A. students who were more demographically different
and groups that were more demographically heterogeneous were less likely to perceive and develop
cooperative norms early in the groups' formation
than more similar individuals and more homogeneous groups, respectively. Similarly, being demographically different was negatively associated
with perceptions of cooperative norms for financial
services firm officers with low (newcomer) rather
than high (veteran) tenure. This pattern of findings
is consistent with prior research showing how
functional antagonism precludes a simultaneous
focus on characteristics that differentiate people,
such as demography, and characteristics that assimilate people, such as group membership (e.g.,
Chatman et al., 1998). Our study adds to existing
research by suggesting that this effect is stronger
early on in both a person's membership in a group
and the group's existence.
For the M.B.A. sample, the connection between
demographic heterogeneity and perceptions of cooperative norms had dissipated completely by the
end of the 15-week project. But the relationship
between being demographically different and perceiving norms as less cooperative endured for at
least 2.6 years for the newcomers among the financial services officers. This longer duration of the
demography-norms link may reflect differences between the samples. We had to assume that newcomers were similar, in many regards, to veterans
when the latter first entered the financial services
firm. In contrast, M.B.A. student teams had fixed
membership for the duration of their projects,
October
which were also highly interdependent tasks that
demanded frequent contact over a concentrated period of time among a small number of people. This
pattern made it possible for members to interact
with one another regularly, which may have accelerated the decreasing impact of demographic heterogeneity on cooperative norms. More realistically, organizations are likely to contain fluid
groups in which employees are members of multiple groups with varying membership at any point
in time. As a result, our findings for the financial
services firm here may be an underestimate of the
effects of tenure on the relationship between demographic heterogeneity and the perceptions of cooperative norms, because we did not assess how long
each respondent had worked with each other member of her or his business unit. More specific information about employee tenure and longitudinal
data about intraunit contact over time would be
useful to gain a finer-grained understanding of the
effects of tenure on the demography-norms link
(e.g., Harrison & Carroll, 1991).
Cooperative norms were more stable over time
among people who were more demographically
similar and less stable among people who were
more different, depending on their contact with
other members. Following contact, different people
may be more likely to recategorize group members
by viewing work group membership as more salient
than previously assigned demographically based
out-group identities. Thus, the test of hypotheses
2a and 2b revealed an important distinction: contact is significantly more influential for demographically different people than for those who are
similar.
We argued that the relatively greater influence of
contact on demographically different people's perceptions of cooperative norms may be explained by
the number of dissimilar relationships being managed. As an individual's relational demography
score increases, he or she has more dyadic relationships with group members who are different from
him or her. Assuming that contact works to improve different people's impressions of each other,
it is only a small step to suggest that demographically different people's impressions of group norms
will improve relatively more because they have
more relationships with other group members that
benefit from increased contact. Interestingly, our
results for Hypothesis 2a actually surpassed our
predictions. People who were more different were,
as predicted, more positively affected by contact
but, by the end of the project, they also perceived
norms as significantly more cooperative than did
similar people who had experienced equivalent
amounts of contact. Given that increased contact
2001
Chatman and Flynn
with different others is likely to be more uncomfortable or taxing, those who are more different may
have escalated their commitment to the group (e.g.,
Staw, 1976), viewing the group's objectives as more
salient and cooperative norms as more prevalent as
a result. Future research might assess the comparative level of discomfort similar and different members experience when interacting with one another.
The Main and Mediating Effects of
Cooperative Norms
Demographic heterogeneity was negatively related to holding early rather than later meetings,
members' satisfaction with their team experience,
individual performance (as rated by supervisors
and as reflected in compensation increases in the
financial services firm), and team efficiency and
effectiveness. We found, in all seven equations
tested in the two studies, that when people perceived or teams emphasized more cooperative
norms, meetings were held earlier rather than later,
members were more satisfied, people performed
better, and teams were more effective.
Most importantly, however, we found consistent
support for the mediating role of cooperative norms
in all five (out of seven) equations that we tested,
though the mediating effect for Hypothesis 6b was
modest. Taken together, these results support a
model in which cooperative norms mediate the
effects of demographic heterogeneity on work processes (Hypotheses 3a and 3b) and outcomes (Hypotheses 4a-6b). This is a critical theoretical
contribution, because it suggests that, in past investigations of teams, researchers may have overemphasized the direct influence of demographic
composition and failed to appreciate the influence
of norms. Indeed, in past research, demographic
differences may have been confounded with cooperative team norms. For example, all groups were
encouraged to adopt cooperative work norms,
which were neither varied nor accounted for in
analyses, in one study (Thornburg, 1991), precluding a determination of whether group heterogeneity
or cooperative norms caused variations in group
effectiveness.
Our mediation findings also imply that research
should focus on the factors that cause specific
group norms to emerge, rather than take for granted
the direct effects of demographic composition on
work processes and outcomes. Team norms can be
identified to predict whether a diverse group (or
any group) will be more or less effective. The tendency for demographically heterogeneous people
and groups to be respectively less likely to perceive
and form cooperative norms early (in their group
971
membership or in the group's existence) is predictable. And, although people who were more different from their coworkers were less satisfied, were
poorer performers, and were paid less than those
who were more similar, these outcomes might have
been improved had they viewed norms as more
cooperative. Likewise, more heterogeneous teams
delayed meetings and were judged to be less efficient and effective. An increased emphasis on cooperative norms may have enhanced their efficiency and effectiveness.
Limitations and Implications for Future Research
One limitation of our first study was a lack of
external validity. We attempted to address this in
several ways. First, we drew participants from an
M.B.A. program that is highly selective, only admitting candidates with substantial work experience. Additionally, the consulting project, which
counted for nearly half of the course grade, focused
on a problem in a real organization. Given their
career objectives, these students tended to be
highly "instrumental" in choosing the organization
and problem on which to focus. Thus, performance
on the project had real consequences because it
could be used as a vehicle for long-term career
objectives. Although we believe that the second
study, of financial services firm officers, provides a
more compelling external validation of the original
study, the M.B.A. student project may begin to
approximate the kinds of stakes and performance
pressures that people face in a real work group
setting.
Future research might also address the salience
and relative influence of other relevant demographic characteristics, such as length of stay and
functional background, as these differences may
have different effects on cooperative norms in different organizations (e.g., Spataro, 2000). Further,
researchers have not consistently agreed on the best
way to represent compositional effects. In this
study, we aggregated three demographic characteristics, but each dimension may be a more or less
potent cause of feeling different.
We used retrospective reports of meeting frequency. The accuracy of respondents' reports may
have been biased by their previous reports of team
norms, because we collected data on team norms
and meeting frequency at time 2 with the same
survey. Future research might use alternative methods of measuring meeting frequency, such as observation or diary accounts during the course of a
project. Further, although the measures collected
here were highly reliable across raters, groups of
raters may be uniformly biased in their recollection
972
Academy
of Management
of meetings and team norms. Further, we assessed
when meetings took place, but not what occurred
in these meetings. Given the importance of contact,
particularly as a way of changing demographically
different people's perceptions of norms, in the future researchers should investigate the content of
group meetings to identify the activities associated
with earlier and later meetings. It may be that earlier meetings are more conducive to considering
tasks at a more leisurely pace, one that is more
likely to stimulate creative, high-quality ideas. Experiencing the pressure of an impending deadline
may make it more difficult for members to think
creatively and develop quality output.
Conclusions
Demography research has produced enough
mixed evidence to support two diametrically opposed positions: the value in diversity hypothesis
(e.g., Cox, et al., 1991) and the conclusion, drawn
by Williams and O'Reilly, that "increased diversity
typically has negative effects on the ability of the
group to meet its members' needs and to function
effectively over time" (1998: 166). One common
feature of both positions is a paucity of research
specifying how and when demographic differences
lead to positive versus negative outcomes. By specifying the time frame in which demographic heterogeneity has the greatest impact- early in a group's
from this study may alert reexistence-findings
searchers to the fleeting nature of the effects of
demographic heterogeneity, per se, on group processes and outcomes. Instead, to understand groups
in organizational settings, more insight may be derived by focusing on the intervening processes, specifically, on how cooperative norms are constructed and the factors leading groups' members to
focus on their shared fate.
Joural
October
Bettenhausen, K. L., & Murnighan, J. K. 1985. The emergence of norms in competitive decision making
groups. Administrative Science Quarterly, 30: 350372.
Bettenhausen, K. L., & Murnighan, J. K. 1991. The development of an intragroup norm and the effects of
interpersonal and structural challenges. Administrative Science Quarterly, 36: 20-35.
Birenbaum, A., & Sagarin, E. 1976. Norms and human
behavior. New York: Praeger.
Brewer, M. B., & Brown, R. J. 1998. Intergroup relations.
In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.),
Handbook of social psychology (4th ed.): 554-594.
Boston: McGraw-Hill.
Brewer, M. B., & Miller, N. 1984. Beyond the contact
hypothesis: Theoretical perspectives on desegregation. In N. Miller & M. Brewer (Eds.), Groups in
conflict: A psychology of desegregation: 281-302.
San Diego: Academic.
Chatman, J., Polzer, J., Barsade, S., & Neale, M. 1998.
Being different yet feeling similar: The influence of
demographic composition and organizational culture on work processes and outcomes. Administrative Science Quarterly, 41: 423.
Cox, T., Lobel, S., & McLeod, P. 1991. Effects of ethnic
group cultural differences on cooperative and competitive behavior on a group task. Academy of Management Journal, 34: 827-847.
Dovidio, J. F., Gaertner, S. L., & Validzic, A. 1998. Intergroup bias: Status, differentiation, and a common
in-group identity. Journal of Personality and Social
Psychology, 75: 109-120.
B. S. 1986. Organizational structure,
problem solving, and effectiveness. San Francisco:
Georgopoulos,
Jossey-Bass.
Gersick, C. 1988. Time and transition in work teams:
Toward a new model of group development. Academy of Management Journal, 31: 9-41.
REFERENCES
Guzzo, R. A., & Dickson, M. W. 1996. Teams in organizations: Recent research on performance and effectiveness. In J. T. Spence, J. M. Barley, &J. Foss (Eds.),
Annual review of psychology, 47: 307-338. Palo
Alto, CA: Annual Reviews.
Alagna, S., Reddy, D., & Collins, D. 1982. Perceptions of
functioning in mixed-sex and male medical training.
groups. lournal of Medical Education, 57: 801- 803.
Hackman, J. R. 1987. The design of work teams. In J.
Lorsch (Ed.), Handbook of organizational behavior: 315-342. Englewood Cliffs, NJ: Prentice-Hall.
Allison, P. D. 1978. Measures of inequality. American
Sociological Review, 43: 865-880.
Harrison, D. A., Price, K. H., & Bell, M. P. 1998. Beyond
relational demography: Time and the effects of surfaceand deep-level diversity on work group cohesion.
Academy of ManagementJouzmal,41: 96-107.
Ancona, D., & Caldwell, D. 1992. Demography and design: Predictors of new product team performance.
Organization Science, 3: 321-341.
Baron, R. M., & Kenny, D. A. 1986. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considand Social
erations. Journal of Personality
Psychology, 51: 1173-1182.
Harrison, J. R., & Carroll, G. R. 1991. Keeping the faith: A
model of cultural transmission in formal organizations. Administrative Science Quarterly, 36: 552582.
Heilman, M. E. 1994. Affirmative action: Some unintended consequences for working women. In B. M.
2001
Chatman and Flynn
Staw & L. L. Cummings (Eds.), Research in organizational behavior, vol. 16: 125-170. Greenwich, CT:
JAI Press.
Hoffman, L. R., Harburg, E., & Maier, N. R. F. 1962.
Differences and disagreement as factors in creative
group problem solving. Journal of Abnormal and
Social Psychology, 64: 206-214.
Islam, M. R., & Hewstone, M. 1993. Dimensions of contact as predictors of intergroup anxiety, perceived
out-group variability, and out-group attitude: An integrative model. Personality and Social Psychology
Bulletin, 19: 700-710.
James, L. R., & Brett, J. M. 1984. Mediators, moderators,
and tests for mediation. Journal of Applied Psychology, 69: 307-321.
James, L. R., Demaree, R. G., & Wolf, G. 1984. Estimating
within-group interrater reliability with and without
response bias. journal of Applied Psychology, 69:
85-98.
Jehn, K. A., Northcraft, G. B., & Neale, M. A. 1999. Why
differences make a difference: A field study of diversity, conflict, and performance in workgroups.
Administrative Science Quarterly, 44: 741-763.
Kenny, D., & LaVoie, L. 1985. Separating individual and
group effects. Journal of Personality and Social
Psychology, 48: 339-348.
Kirchmeyer, C., & Cohen, A. 1992. Multicultural groups:
Their performance and reactions with constructive
conflict. Group & Organization Management, 17:
153-170.
Lawrence, B. 1997. The black box of organizational demography. Organization Science, 8: 1-22.
973
Pfeffer, J. 1983. Organizational demography. In L. L.
Cummings & B. M. Staw (Eds.), Research in organizational behavior, vol. 3: 1-52. Greenwich, CT: JAI
Press.
Riordan, C. M., & Shore, L. M. 1997. Demographic diversity and employee attitudes: An empirical examination of relational demography within work units.
Journal of Applied Psychology, 82: 342-358.
Schoonhoven, C. B. 1981. Problems with contingency
theory: Testing assumptions hidden within the language of contingency "theory." Administrative Science Quarterly, 26: 349-377.
Sherif, M. H., Harvey, 0. J., White, B. J., Hood, W. R., &
Sherif, C. 1954. Experimental study of positive and
negative intergroup attitudes between experimentally produced groups: Robbers Cave experiment.
Norman: University Oklahoma.
Simons, T. 1995. Top management team consensus, heterogeneity, and debate as contingent predictors of
company performance: The complimentarity of
group structure and process. Academy of Management Best Paper Proceedings: 62-66.
Spataro, S. 2000. Not all differences are the same: The
role of status in predicting reactions to demographic diversity in organizations. Unpublished
doctoral dissertation, University of California,
Berkeley.
Staw, B. M. 1977. Knee-deep in the Big Muddy: A study
of escalating commitment to a chosen course of action. Organizational Behavior & Human Decision
Processes, 16: 27-44.
Stephan, W. G., & Brigham, J. C. (Eds.). 1985. Intergroup
contact. Journal of Social Issues, 41(3): 1-179.
Levine, J. M., & Moreland, R. L. 1991. Culture and socialization in work groups. In L. Resnick & J. Levine
(Eds.), Perspectives on socially shared cognition:
257-279. Washington, DC: APA.
Stroessner, S. J. 1996. Social categorization by race or
sex: Effects of perceived non-normalcy on response
times. Social Cognition, 14: 274-276.
Markus, H. R., & Cross, S. 1990. The interpersonal self. In
L. A. Pervin (Ed.), Handbook of personality: Theory
and research: 576-608. New York: Gilford Press.
Thornburg, T. H. 1991. Group size and member diversity
influence on creative performance. Journal of Creative Behavior, 25: 324-333.
Messick, D., & Mackie, D. 1989. Intergroup relations. In
M. R. Rosenzweig & L. W. Porter (Eds.), Annual
review of psychology, vol. 40: 45-81. Palo Alto, CA:
Annual Reviews.
Tsui, A. S., Egan, T. D., & O'Reilly, C. 1992. Being different: Relational demography and organizational attachment. Administrative Science Quarterly, 37:
549-579.
O'Reilly, C. A., Caldwell, D. F., & Barnett, W. P. 1989.
Work group demography, social integration, and
turnover. Administrative Science Quarterly, 34:
21-37.
Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, C.
1994. Self and collective: Cognition and social context. Personality and Social Psychology Bulletin,
20: 454-463.
O'Reilly, C. A., & Flatt, S. 1989. Executive team demog-
Wageman, R. 1995. Interdependence and group effectiveness. Administrative Science Quarterly, 40: 145180.
raphy, organizational innovation, andfirm performance. Paper presented at the annual meeting of the
Academy of Management, Washington, DC.
O'Reilly, C. A., Williams, K., & Barsade, S. 1997. Group
demography and innovation: Does diversity help? In
E. Mannix & M. Neale (Eds.), Research in the management of groups and teams, vol. 1: 183-207.
Greenwich, CT: JAI Press.
Wagner, J. A. 1995. Studies of individualism-collectivism: Effects on cooperation in groups. Academy of
Management journal, 38: 152-172.
Watson, W. E., Kumar, K., & Michaelsen, L. K. 1993.
Cultural diversity's impact on interaction process
and performance: Comparing homogeneous and di-
974
Academy of Management Joural
October
Jour-
JenniferA. Chatman (chatman@hass.berkeley.edu) is the
Wayne, S., & Liden, R. 1995. Effects of impression management on performance ratings: A longitudinal study.
Academy of Management Journal, 38: 232-260.
Harold Furst Professor of Management Philosophy and
Values at the University of California's Haas School of
Business. Her research interests focus on organizational
culture, cooperation, person-culture fit, and demography. She received her Ph.D. in organizational behavior
from the University of California, Berkeley.
verse task groups. Academy of Management
nal, 36: 590-602.
Williams, K. Y., & O'Reilly, C. A. 1998. Forty years of
diversity research: A review. In B. M. Staw & L. L.
Cummings (Eds.), Research in organizational behavior, vol. 20: 33-140. Greenwich, CT: JAI Press.
Zenger, T. R., & Lawrence, B. S. 1989. Organizational
demography: The differential effects of age and tenure distributions on technical communication.
Academy of Management Journal, 32: 353-376.
Francis J. Flynn is an assistant professor of management
at the Columbia Business School, Columbia University.
He earned his Ph.D. in organizational behavior from the
University of California, Berkeley. His research interests
focus on the determinants and consequences of social
influence in organizations.